Dynamic modeling of local reaction conditions in an agitated aerobic fermenter

1
Dynamic modeling of local reaction conditions
in an agitated aerobic fermenter
NAMF Mixing XX, Parksville, Vancouver Island,
Canada
June 26 July 1, 2005

• Marko Laakkonen, Pasi Moilanen, Ville Alopaeus,
  Juhani Aittamaa
• Helsinki University of Technology
• Laboratory of Chemical Engineering
• Finland

2
Motivation

• Problems of aerobic
• fermenter design
• Gas-liquid mass transfer
• limitations
• Mixture inhomogeneity
• Changing physical properties
• CFD tools too slow for
• the simulation of a long
• fermentation batch

3
The objectives

• To develop a bioreactor model for the
  investigation of
• Batch fermentation dynamics
• Local reaction and mass transfer conditions
• To validate the submodels against stirred tank
  experiments with xanthan solutions

xanthan 0.13 w-
0.25 w-
0.50 w-
4
Experiments in 200 dm3 vessel

• xanthan solutions 0 2.5
  w-
• Gassing 0.1 1.0 vvm
• Agitation 0.1 3 W/kg
• The measured quantities
• Gassed power consumption
• Overall gas holdup
• Bubble size distributions
• Oxygen mass transfer
• Viscosity measurements

5
Multiblock models for the laboratory vessel and a
0.64 m3 pilot fermenter

• The change of flow patterns from CFD simulations
  at various xanthan concentrations
• Population balances for bubbles
• Multicomponent gas-liquid mass transfer model
• Xanthan fermentation kinetics of Garcia-Ochoa et
  al. (2000)

pilot fermenter
6
Viscosity of xanthan solution

• The model of Carreau (1972)
• where

7
Power consumption of mixing, 200 dm3 vessel
Predicted vs. measured gassed power number

• Ungassed power number
• Gassed power uptake (Cui et
  al. 1996)

8
Population balance for bubbles
dY/dt Slip/convection Breakage
Coalescence_at_ Growth

• Bubble drag (Tzounakos et al. 2004)
• Breakage rates (Luo Svendsen (1996)
• Daughter bubble size distribution (Lehr et
  al. 2002)
• _at_ Coalescence rates (Coulaloglou Tavlarides
  (1977)
• Coalescence efficiencies (Chesters 1991)

9
Bubble size distributions, 200 dm3
vessel
Markers measured Lines simulated
Agitation 390 rpm Gassing 0.5 vvm
10
0.25 w- xanthan, 200 dm3 vessel
390 rpm 0.5 vvm
Measured
Simulated
11
Gas holdup, 200 dm3 vessel
Measured
12
Gas-liquid mass transfer fluxes

• Two-film theory with
  a simplified
  solution of
  Maxwell-Stefan diffusion
• Liquid side film coefficients
  (Kawase et
  al. 1992)
• Gas side film coefficients A rational
  approximation for the diffusion in bubbles
  (Alopaeus, 2001)
• Equilibrium from Henrys law with salting out
  correction (Rischbieter Schumpe, 1996)

13
Measured vs. simulated kLa 200 dm3 vessel
_at_
Multiblock simulation with population balance
and mass transfer models _at_An empirical
correlation for xanthan fermentation broth
14
Fermenter simulations

• Chemical compounds in gas and liquid
• H2O, CO2, O2 and N2
• Reacting compounds in liquid
• Biomass, Xanthan, Nutrient, Carbon source
• 40 bubble size classes with adaptive
  discretization
• Simulation cases
• S1 Constant stirring speed 300 rpm (1 W/kg)
• S2 Stirring speeds 300 ? 475 rpm (1 4 W/kg)

15
Overall performance of fermenter
16
Inhomogeneity of xanthan reaction
Spatial autocorrelation (GR) Magnitude of
local gradients
Spatial standard deviation (SD)
17
Mean bubble sizes d32 (Sd3/Sd2)
S2, t 24 h
S1, t 65 h
S1, t 20 h
300 rpm
475 rpm
300 rpm
2.2 w- xanthan
1.2 w- xanthan
2.4 w- xanthan
18
Mass transfer coefficients (kLa)
S2, t 24 h
S1, t 65 h
S1, t 20 h
300 rpm
300 rpm
475 rpm
1.2 w- xanthan
2.2w- xanthan
2.4 w- xanthan
19
Dissolved oxygen
S2, t 24 h
S1, t 65 h
S1, t 20 h
300 rpm
300 rpm
475 rpm
1.2 w- xanthan
2.2w- xanthan
2.4 w- xanthan
20
Xanthan reaction rates
S1, t 20 h
S1, t 65 h
S2, t 24 h
300 rpm
300 rpm
475 rpm
1.2 w- xanthan
2.4 w- xanthan
2.2 w- xanthan
21
Conclusions

• Complex, non-newtonian gas-liquid hydrodynamics
  and mass transfer were described succesfully for
  the laboratory stirred tank
• The developed bioreactor model can be used to
  investigate the dynamics of local mass
  transfer and reaction rates
  in agitated aerobic fermenters

22
Acknowledgement

• Asta Nurmela, Suvi Jussila and Elina Nauha for
  their contribution to the experimental part of
  work
• Financial support from the Graduate School of
  Chemical Engineering, Neobio (New design tool for
  bioreactors) and Modcher (Modelling of Chemical
  Reactors) projects funded by the National
  Technology Agency of Finland (TEKES)